|
|
All knowledge is of an approximate character and always will be (B.
Russell, Human Knowledge, 1948, pg 497 and 507). Our formalisms
abstract, idealize, and simplify (R. L. Epstein, Propositional Logics,
2001, Ch XI and E. Bender, An Intro. to Math. Modeling, 1978, pg v and
2). Each formalism is an idealization, often times approximating in
its own DIFFERENT ways, each offering somewhat different coverage of
the domain. Having MULTIPLE overlaping theories of a knowledge domain
is then better than having just one theory (R. Jones, APS general
meeting, April 2004). Theories are not unique (T. M. Mitchell,
Machine Learning, 1997, pg 65-66 and Cooper, Machine Learning, vol. 9,
1192, pg 319). In the future every field will possess multiple
theories of its domain and scientific work and engineering will be
performed based on the ensemble predictions of ALL of these. In some
cases the theories may be quite divergent, differing greatly one from
the other. This idea can be considered an extension of Bohr's notion
of complementarity, "...different experimental
arrangements...described by different physical concepts...together and
only together exhaust the definable information we can obtain about
the object." (H. J. Folse, The Philosophy of Neils Bohr, 1985, pg
238). Although finding the "correct" or "most probable" theory has
been the goal of scientific investigation in the past we now know that
the pluralistic science that I am describing here is more successful
(Peter Cheeseman in The Mathematics of Generalization, D. H. Wolpert,
ed., 1995,
pg. 315). This is not postmodernism. Science is not democratic.
Theories are accepted based upon experimental evidence not human
opinion or preference.
In keeping with the new pluralism when I was doing magnetic fusion
energy research it was important that I studied open traps (Plasma
Phys, vol. 22, pg 753, 1980) as well as closed magnetic systems (Can.
J. Phys., vol. 57, pg 635, 1979). And in studying open systems it was
significant that I explored both adiabatic (Nuovo Cimento vol. 9D, pg
247, 1987) and nonadiabatic traps (Proc. 10th Sym. Fusion Eng., pg
1864, 1984). It was equally important that I study alternatives to
magnetic confinement (Pramana, vol. 20, pg 47, 1983). When I turned
to artificial intelligence work I pursued neural networks (Trans.
Kansas Acad. Sci., vol. 99, pg 85, 1997) as well as traditional logic
based systems (Trans. Kansas Acad. Sci., vol. 102, pg 32, 1999).
R. Jones
Prof. of Physical Science |
|
|
